Problem

Show that $V(X) =0$ if and only if there is a constanct $c$ such that $P(X = c) = 1$.

Solution

First, let's note that since we are talking about a discrete $P(X=c)$, $X$ is a discrete r.v. Let's assign these two conditions.

  • a: $V(X) = 0$
  • b: $P(X=c) = 1$

Part 1: show $b \implies a$

$P(X=c) = 1$ aka $f_X(c) = 1$. Recall that from the basics of probability distributions $\sum_i f_X(i) = 1$ and that every $f_X(i) \geq 0$. If there exists a $c$ such that $f_X(c) = 1$ then every other $f_X(i)$ must be 0.

$E(X) = \mu = f_X(c)c = 1 \times c = c$

$E(X^2) = \sum_i x^2 f_X(x) = c^2 f_x(c) = c^2$

So $V(X) = E(X^2) - \mu^2 = c^2 - c^2 = 0$.

Part 2 (TODO): show $a \implies b$

I got stumped by this only to find that the solution used Chebyshev's inequality, which isn't covered until chapter 4. Screw this!

Sources

  • 36-705 hw1 problem 8